基于指数平滑法和ARIMA的交通量组合预测模型应用研究

发布时间：2019-04-19 04:41

【摘要】：道路交通流预测是现代交通管理和规划的重要任务,因此准确地预测交通流的未来趋势也是非常重要的,这就需要研究交通量预测的方法,提高预测精度,从而保证待建交通项目决策的科学性和合理性,进而减少道路规划设计过程中的资源浪费,提高社会效益。通过广泛收集相关文献资料,结合小波分析和指数平滑法研究的成果,论文对现有交通量预测方法进行了分析,论述了国内外交通预测理论的发展和应用,分析了灰色预测理论、遗传算法、神经网络等的优点和不足,在此基础上,结合交通量历史数据的随机性与非线性变化等特点,通过分析数据信号采集中出现的异常数据,探讨了异常数据的识别和修正的不同方法,并实际应用了这些方法进行交通量数据序列中异常数据的修正。另外,论文又进行了实证研究,结合郑少高速公路道路交通量及其相关数据资料,通过对比分析,进一步验证预测方法的可行性和预测结果的可靠性。根据指数平滑法的计算原理,并结合郑少高速公路交通量数据,论文通过把不同尺度上的因素分离开来(交通量受很多因素影响),使规律性更加明显从而更易于预测。本论文研究将小波方法引入到交通量预测中,利用小波多尺度分析的功能和三次指数平滑法的预测,提出了小波多尺度指数平滑复合预测法,将小波多尺度功能分析原理与指数平滑法有机地结合到一起,同时采用ARIMA一系列模型通过Spss软件对交通量进行预测,从本质上把握交通量的非线性变化特征。论文通过进一步研究和应用组合预测方法,把上述两种方法通过线性组合有效地结合起来,通过实例很好地验证了小波多尺度指数平滑预测和ARIMA模型组合预测模型预测交通量,两种方法的组合预测效果良好,较为准确地预测了郑少高速未来交通量长期趋势,丰富和发展了交通量预测方法,对高速公路的项目决策与运营管理,具有一定的重要学术研究价值和工程应用价值。
[Abstract]:Road traffic flow forecasting is an important task in modern traffic management and planning, so it is also very important to accurately predict the future trend of traffic flow, so it is necessary to study the methods of traffic volume prediction and improve the prediction accuracy. In order to ensure the scientific and rational decision-making of traffic projects to be built, the waste of resources in the process of road planning and design is reduced, and the social benefit is improved. Through the extensive collection of relevant literature, combined with the research results of wavelet analysis and exponential smoothing method, this paper analyzes the existing traffic forecasting methods, and discusses the development and application of traffic forecasting theory at home and abroad. This paper analyzes the advantages and disadvantages of grey prediction theory, genetic algorithm, neural network, etc. On this basis, combining with the randomness and non-linear change of traffic volume historical data, the abnormal data appearing in data signal collection are analyzed. In this paper, different methods for identifying and correcting abnormal data are discussed, and these methods are applied to correct abnormal data in traffic data series. In addition, the paper also carries on the empirical research, combines the Zheng-Xiao expressway road traffic volume and the related data, through the comparative analysis, further validates the feasibility of the forecast method and the reliability of the forecast result. According to the calculation principle of exponential smoothing method and combined with the traffic volume data of Zheng-Shao Expressway, the paper makes the regularity more obvious and easy to predict by separating the factors on different scales (the traffic volume is influenced by many factors). In this paper, wavelet method is introduced into traffic volume prediction. The wavelet multi-scale exponential smoothing method is proposed by using the function of wavelet multi-scale analysis and the prediction of cubic exponential smoothing method. The principle of wavelet multi-scale function analysis and exponential smoothing method are combined organically. At the same time, a series of ARIMA models are used to predict traffic volume through Spss software to grasp the nonlinear characteristics of traffic volume in essence. Through further research and application of combinatorial forecasting method, this paper combines the above two methods effectively through linear combination, and validates the prediction of traffic volume by wavelet multi-scale exponential smoothing prediction and ARIMA model combination forecasting model. The combination of the two methods can predict the long-term trend of the future traffic volume of Zheng-Xiao Expressway accurately, enrich and develop the forecasting method of the traffic volume, and make decision-making and operation management of the expressway project. It has certain important academic research value and engineering application value.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.14

【相似文献】